A third species of Polyspatha, an Africanendemic genus of Commelinaceae

Abstract Polyspatha oligospatha Faden, the third species in a small African endemic genus of Commelinaceae, is described. It is widespread but has been overlooked because of its small stature and resemblance to small plants of Polyspatha paniculata. It differs from both Polyspatha paniculata and Polyspatha hirsuta, the two other species, by its leaf pubescence, fewer, more widely spaced and usually patent spathes, deeply ridged seeds with numerous knobby, transversely interrupted ridges, and morning anthesis. It occurs throughout the Congolian forests from Cameroon to Uganda, but it is also disjunct in Ivory Coast, across the Dahomey gap.


Introduction
The African endemic genus Polyspatha (Commelinaceae) has generally been considered to consist of two species (Morton 1967;Brenan 1968). It was described by Bentham (1849) for a single West African species, P. paniculata Benth. In his monograph of the family, Clarke (1881) also recognized P. paniculata var. glaucescens C. B. Clarke, which he subsequently abandoned in the Flora of Tropical Africa (Clarke 1901). Clarke's variety was raised to species rank as P. glaucescens (C. B. Clarke) Hutch. in the first edition of the Flora of West Tropical Africa (Hutchinson 1936) but the name was abandoned by Morton in his survey of Commelinaceae of West Africa (Morton 1967) and treated in synonymy under P. paniculata in the second edition of the Flora of West Tropical Africa (Brenan 1968). The second currently recognized species of Polyspatha, P. hirsuta Mildbr., was described by Mildbraed (1925). Faden (1998) recorded Polyspatha as having three species, but the third species had not been described.
When I first began working on the Commelinaceae of Cameroon and examined the Polyspatha specimens at the Herbier National du Cameroun (YA) in Yaoundé, Cameroon in 1986, it appeared that there were three distinct species of Polyspatha, not the usually accepted two. I came to the same conclusion from studies at the Muséum National d'Histoire Naturelle, Paris (P) the following year. Later studies at the Royal Botanic Gardens Kew (K), in connection with work on the Flora of Tropical East Africa, confirmed that the same three species also occurred in Uganda, some 2000 km away. Studies of collections from other herbaria, particularly the National Botanic Garden of Belgium, Brussels (BR), confirmed that the three species also were present in the intervening territory. It was particularly significant that this species was also found to occur in Ivory Coast, almost 1900 km from the nearest locality in Cameroon.

Methods
More than 540 herbarium specimens of Polyspatha were studied from the following institutions: all the collections from BM, BR, EA, K, NY, MO, US, WAG and YA, and some of the collections from P. Living material of Polyspatha oligospatha (Poulsen 1275, originally from Uganda, and Faden et al. 86/2, originally from Cameroon) was cultivated at the Smithsonian Institution Botany Research Greenhouse, in Suitland, Maryland, USA and provided data on flowering times, root tips for chromosome counts, seed collections and liquid preserved flowers. A mitotic chromosome count was obtained using the techniques of Faden and Suda (1980). The seeds were photographed using an Olympus SZX-12MDU dissecting microscope, equipped with an Olympus Q-Color 5 camera, using Image-Pro and extended depth of field (EDF). The distribution map was created using ArcGIS once latitudes and longitudes were determined or estimated for the collections seen.  (Mabira), September 1922, R. A. Dummer 5531 (US!, holo, K!, iso).

Discussion
Polyspatha oligospatha [etymology: many spathes, few spathes] differs from P. paniculata by the consistent presence of long, uniseriate hairs on the adaxial leaf surface (vs. none), which is never scabrous (vs. often scabrous) and by the usually fewer, smaller, blunter, more widely spaced and less deflexed spathes. It differs from P. hirsuta by the usually sparser and shorter uniseriate hairs on the adaxial leaf surface, the absence of such hairs from the abaxial leaf surface and usually the internodes and sheaths (except for the sheath summit), and the inflorescence consisting of fewer, less congested spathes with shorter marginal hairs and with lustrous, brown, bead-like cells on the entire spathe surface or mainly near the midrib (vs. such cells completely lacking).
The new species differs from both other species by its generally smaller stature (but see below) and seeds with numerous, knobby, transversely interrupted ridges as compared with generally fewer, smoother, uninterrupted ridges in the seeds of P. paniculata and testa shallowly ridged-reticulate in P. hirsuta (Figure 1: 7, 8; Plate 1). Leaf pubescence in P. hirsuta can be quite variable, but plants with sparse pubescence or with long hairs confined to the adaxial leaf surface are restricted to Nigeria and further west, overlapping with P. oligospatha only in Ivory Coast. Thus throughout nearly all of the range of P. oligospatha, the pubescence differences of the leaf lamina, sheath, and internodes between P. oligospatha and P. hirsuta are consistent. The often much less pubescent plants of West African P. hirsuta still exhibit the stature, leaf shape and spathes that are typical of the species, so they would not be readily confused with P. oligospatha even if the two species were more broadly sympatric.
The new species has been confused in the past with small specimens of P. paniculata which can be of similar stature and appearance. Some may even have very similar looking spathes and smaller capsules and seeds than in typical plants of P. paniculata. Such plants can be separated from P. oligospatha by the absence of long, uniseriate hairs from the leaves, the spathes generally more crowded, overlapping and deflexed against the inflorescence axis, and the seeds with generally fewer, smooth, uninterrupted ridges, as in typical P. paniculata (Plate 1-1).
This species has been overlooked or misinterpreted partly because of Clarke's P. paniculata var. glaucescens, which was raised to a species in the first edition of Flora of West Tropical Africa (FWTA) by Hutchinson (1936), although it had already been abandoned by Clarke (1901) in Floral of Tropical Africa. From a study of the type of this variety, Brenan (1968) in the second edtion of FWTA correctly concluded that it was no more than a depauperate form of P. paniculata, agreeing with Morton (1967). This led Brenan to annotate the only specimen at Kew of P. oligospatha from West Tropical Africa, Aké Assi 5707 from Ivory Coast, as "small form--I do not consider P. glaucescens...as distinct." On the other hand, it is clear that Brenan had begun to suspect that there was more variation in the genus than the two West African species by his note on the species folder of P. paniculata from Uganda, which included specimens of P. oligospatha: "Some of these Uganda sheets look distinct and should be investigated." The collection Letouzey 11465 (Figure 2, arrow) from Cameroon is accepted as P. oligospatha with reservation. The pubescence of the upper leaf surface is more like that of P. hirsuta, with the uniseriate hairs longer, denser and paler than in typical P. oligospatha. However the absence of uniseriate hairs from the lower leaf surface, sheaths (except for the summits) and internodes distinguishes it from all Cameroonian specimens of P. hirsuta. Moreover, the outer surface of the small spathes is composed solely of lustrous, dark brown, bead-like cells, which are present in all specimens of P. oligospatha but none of P. hirsuta. The short marginal hairs on the spathes also agree with P. oligospatha, but are unusual for P. hirsuta in the same region. Overall, Letouzey 11465 is best treated as an atypical specimen of P. oligospatha.
Three sheets of Letouzey 11465 have been seen but those at P and YA were not critically examined. Thus my observations and conclusions have been based solely on the duplicate of this collection from BR. Unfortunately, it lacks seeds, which would be diagnostic.
Manning 2166 is unusual for P. oligospatha in several characters. The internodes and surface of the leaf sheaths bear numerous uniseriate hairs, a pubescence that has not been noted in any other collection of P. oligospatha. The spathes are also more pubescent than usual, bearing many, long, uniseriate hairs on the surface as well as along the midrib. The longest marginal hairs are >1.5 mm long, the normal maximum for P. oligospatha. Moreover, the spathes have very few lustrous, brown, bead-like cells and these are confined to the midrib area towards the spathe bases. In all these characters Manning 2166 resembles P. hirsuta. It differs from that species by its shorter stature, absence of long hairs from the abaxial leaf surface--they are typically present in Cameroonian collections--shorter uniseriate hairs on the internodes and sheath surfaces, and more spaced, patent and pointed spathes that bear at least a small number lustrous, brown, bead-like cells along the midrib. Manning 2166 might possibly represent a hybrid of P. oligospatha and P. hirsuta, but in view of the different times of anthesis in these species, I think it is best considered an atypical specimen of P. oligospatha. As in the case of Letouzey 11465, only the WAG sheet of Manning 2166, of the three specimens seen, was studied in great detail, thus reinterpretation might be necessary when the other sheets can be reexamined.
Polyspatha oligospatha has sometimes been collected as part of mixed collections with P. paniculata. I have seen five such mixed collections: Lieberberg 922 & 922A(K) from Uganda, Seret 70 (BR) from the Democratic Republic of Congo, Schlechter 12714 (BR, K) from the Republic of Congo, and Letouzey 8742 from Cameroon. Possibly when further duplicates of already seen collections are examined in additional institutions, other mixed collections will be discovered. Thus far no mixed collections have been seen of P. oligospatha and P. hirsuta.
The disjunct occurrence of P. oligospatha in Ivory Coast is probably not an artifact of under-collection because in both Ghana and Nigeria, the two larger of the four countries between Ivory Coast and Cameroon, there were collectors who took a great interest in Commelinaceae. J. K. Morton made numerous collections of Commelinaceae in Ghana and elsewhere and wrote several important papers on the family. D. P. Stanfield was an avid collector of Commelinaceae in Nigeria and was rewarded for his efforts and lengthy and detailed notes by having a new genus Stanfieldiella named for him by Brenan (1960). Thus although P. oligospatha is somewhat inconspicuous I would have expected it to have been collected in Ghana and/or Nigeria if it occurred there.
Another factor that makes this disjunct distribution interesting is that it places Polyspatha oligospatha on both sides of the Dahomey gap, the natural break in the rainforest that separates the Upper Guinean and Congolian forest blocs. A similar distribution pattern is shown by another Commelinaceae species, Palisota ambigua (P.Beauv.) C.B.Clarke, which Flora of West Tropical Africa (Brenan 1968) places in southern Nigeria but no further west. However, I observed four collections of this species from Ivory Coast at P in 1987, including Jolly 39, so it too spans the Dahomey gap. The question remains whether these Ivory Coast occurrences are relicts of a former, continuous distribution or represent more recent long-distance dispersal events. The berry fruits of the Palisota would obviously suggest the latter, but because Polyspatha seeds have no obvious means of dispersal-either short distance or long distance-it is more difficult to use the same argument for that disjunction.
Polyspatha oligospatha appears to be reproductively isolated from the other two species because of its morning anthesis and flowers fading about noon. In Flora of West Tropical Africa Brenan (1968) records the flowering times of P. paniculata as 2:30-5:00 pm and P. hirsuta as 2:30 until evening. Thus anthesis for P. oligospatha does not overlap with those of the other two species.

Conclusion
Polyspatha oligospatha Faden, a new species described herein, is a widespread African rainforest species that has been overlooked but is readily distinguished from the other two species by its seed testa pattern, leaf pubescence, spathe characters and morning flowering.